Serological Response to BNT162b2 Anti-SARS-CoV-2 Vaccination in Patients with Inflammatory Rheumatic Diseases: Results From the RHEUVAX Cohort

Objective In the light of the current COVID-19 epidemic and the availability of effective vaccines, this study aims to identify factors associated with non-response to anti-SARS-CoV-2 vaccines as immunological alteration associated with immune rheumatic diseases (IRD) and immunosuppressive medications may impair the response to vaccination. Methods Volunteers in the health profession community with IRD, age, and sex-matched controls (CTRL) who underwent vaccination with two doses of BNT162b2 were recruited for this study. Anti-Trimeric Spike protein antibodies were assayed eight ± one weeks after the second vaccine dose. Univariate and logistic regression analyses were performed to identify factors independently associated with non-response and low antibody titers. Results Samples were obtained from 237 IRD patients (m/f 73/164, mean age 57, CI 95% [56-59]): 4 autoinflammatory diseases (AI), 62 connective tissue diseases (CTD), 86 rheumatoid arthritis (RA), 71 spondylarthritis (SpA) and 14 vasculitis (Vsc). 232 CTRL were recruited (m/f 71/161, mean age 57, CI 95% [56-58]). Globally, IRD had a lower seroconversion rate (88.6% vs 99.6%, CI 95% OR [1.61-5.73], p<0.001) and lower antibody titer compared to controls (median (IQR) 403 (131.5-1012) versus 1160 (702.5-1675), p<0.001). After logistic regression, age, corticosteroid (CCS), Abatacept and Mycophenolate Mofetil (MMF) use were associated with non-response. Lower antibody titer was associated with the use of MMF, ABA, CCS, Rituximab, tumor necrosis factor inhibitor, JAK inhibitors, and higher age. Conclusion The response to anti-SARS-CoV-2 vaccines is often impaired in IRD patients under treatment and may pose them at higher risk of severe COVID-19. Specific vaccination protocols are desirable for these patients.


INTRODUCTION
It is widely accepted that patients with autoimmune inflammatory rheumatic disease are at higher risk of infection, partially due to treatment (1). In the initial phases of the SARS-CoV-2 pandemic, patients affected by autoimmune and inflammatory rheumatic diseases (IRD) raised concerns about their potentialy higher risk of getting infected and developing severe COVID-19. Several controversial data were made available during the last two years (2). To date, globally considered, the literature suggests that IRD are at higher risk of infection and burdened by higher mortality due to COVID-19 than the general population (3).
Comorbidities associated with higher risk for severe COVID-19 and death are frequent in some IRD patients, such as inflammatory lung disease, cardiovascular disease, and chronic kidney disease. Although not particularly strong at population levels, the risk for a worse prognosis can be exceptionally high in some patients (2,4).
Immediately after the publication of the trials demonstrating the efficacy of the SARS-CoV-2 vaccine in mounting immunity against SARS-CoV-2 and preventing a severe form of COVID-19 and death, the need for prioritization of IRD patients to shield them from SARS-CoV-2 infection appeared evident. Rheumatologists and the scientific community have developed a strong awareness of the importance of vaccination in IRD patients and recommendations regarding the type of vaccines to be used and the timing concerning disease activity and treatment are available.
Vaccination against SARS-CoV-2 was demonstrated to be effective in IRD patient, by reducing the mortality and risk of hospitalization (5). However, a large body of evidence is accumulating on the impact of rheumatic disease and immunosuppressive treatment in response to vaccination, demonstrating a reduction in the immune response for some medication or suggesting a reduction in the titer for others (2).
While evidence from systematic reviews on other vaccines shows some impact of corticosteroids and csDMARDs, such as Methotrexate (MTX), on vaccine immunogenicity, data regarding some bDMARDs and small molecules, such as abatacept (ABA) and JAK inhibitors (JAKi), are now available but limited (6)(7)(8).
The European Alliance of Association for Rheumatology (EULAR) task force was of the opinion that the data on anti-CD20 therapy are most compelling, followed by data on mycophenolate mofetil (MMF) and glucocorticoids (9). Data on methotrexate, JAKi, and abatacept were judged not yet consistent/robust (9). Meanwhile, more data for ABA and JAKi are coming to light from studies with sample sizes ranging to 11 to 54 patients demonstrating reduction in antibody titers (10)(11)(12)(13).
In order to contribute to the current knowledge regarding the impact of IRD and treatment on the response to SARS-CoV-2, we present here the result of an observational cross-sectional study evaluating the serological response and the persistence of antibodies at eight weeks in IRD patient cohort and non-IRD patient controls.

Patient and Control Population Recruitment
In line with the local recommendations, patients affected by IRD in follow-up in the Unit of Rheumatology of the University Hospital "L. Vanvitelli" were invited to receive the vaccination within the same center. From April 1, 2021 to June 30, 2021, 699 IRD outpatients received two doses of the BNT162b2 mRNA anti-SARS-CoV-2 vaccine, three weeks apart. After the approval of an internal ethics board, the possibility to perform a serological test to quantify the antibody response against SARS-CoV-2 was offered to patients eight weeks later. Informed consent was acquired, and serum and clinical data were collected. A population of sex-and age-matched healthy controls belonging to the hospital's clinical and nonclinical personnel was used as control. Inclusion criteria were having a diagnosis of rheumatic disease and having received BNT162b2 vaccine 8 weeks before the sample collection. For the controls, only individuals with a diagnosis of rheumatic disease, immunodeficiency or under immunosuppressive treatment were excluded. At the time of sample collection, history of past-COVID-19, pharmacologic history, cancer history, and information regarding the disease duration were collected. Patients and the public were not involved in this research's design, reporting, and dissemination.

Anti-SARS-CoV-2 Testing
The collected sera from both patients and controls were analyzed in batch after being stored at -80°C using Liaison ® SARS-CoV-2 TrimericS IgG CLIA kits provided by DiaSorin ® following manufacturer protocol. The levels of anti-SARS-CoV-2 IgG antibodies were expressed in World Health Organization International Standard (NIBSC code. 20/268) binding arbitrary unit (BAU/ml). Samples with values of ≥33.8 BAU/ml were considered positive. Patients and the public were not involved in this research's design, reporting, and dissemination.

Statistical Analysis
Depending on their distribution, variables are reported as medians and IQRs or means and standard deviation (SD). Participants were stratified into responders and nonresponders according to Binding Antibody Unit/mL level using 33.8 BAU/mL as the cut-off. For the analysis, the antibody titer was log-transformed to improve normality. For categorical variables, including seroconversion rate, the differences in frequency were tested using c2 and Fisher's exact test. Factors possibly associated with seroconversion were then tested in a binary logistic model using backward stepwise selection (p cutoff for elimination 0.2).
Differences in titer were tested by Mann-Whitney-U test for two-group comparison or Kruskal-Wallis test for three or more comparisons. The effect of demographic, diagnosis, and treatment of the antibody titers was further tested in a linear regression model using backward stepwise selection (p-value cutoff for elimination 0.2) Statistical analysis and graphical presentation of the data were performed using GraphPad Prism (V.9.1.0), IBM SPSS Statistics (V.26), and RStudio (V. 2021.9.1.372) were used for the statistics and graphical presentation of the data.
The demographic characteristics are shown in Table 1. The sex and age-matched control cohort was composed of 232 individuals, mean age of 57 CI 95% [56-58] with an M: F ratio of 71:161.

Seroconversion Rates in IRD Patients Compared to Controls
In the control population, 99.6% of subjects achieved seroconversion while 88.6% of patients with IRD showed above-threshold anti-spike protein antibody titer, 95% CI of Ln (OR) of non-response [1.61-5.73], p < 0.001. The difference was also maintained after stratification by age and sex, Conversely, sex and age did not influence the seroconversion rate within groups ( Figure 1A).

Impact of Treatment on Seroconversion
Similarly, both treatment with csDMARDs and bDMARDs/ tsDMARDs were associated with a lower seroconversion rate   Figure 1E). Independent of the doses, corticosteroid therapy (CCS) was associated with a higher rate of non-response and only 76.9% of patients developed antibodies [Ln (OR) of non-response (2.46-6.62), and this proportion was smaller than CTRL and IRD groups not taking CCS (93.9%), p < 0.001] ( Figure 1F).

Seroconversion in IRD Patients With Past Sars-CoV-2 Infection
Finally, out of 237 patients, only 13 had a documented past Sars-CoV-2 infection and all the patients maintained/developed a serological response after the vaccination ( Figure 1G). Detailed response percentage and comparison are shown in Supplementary Table 1 Binary Logistic Regression Model In order to account for confounding variables, a binary logistic regression model was implemented using the data of the 224 IRD patients who had no evidence of prior infection to investigate associations with antibody positivity. After backward stepwise regression only, MMF, ABA, CCS and age remained significant associated factors (at p<0.05 level) of serological response ( Figure 1H). All but MTX were negatively associated with the seroconversion, corrected ln (OR) 95% CI 0.03-0.08 for MMF, 0.01-0.19 for ABA, 0.06-0.54 for CCS, and 0.89-0.99 for age. Suggesting that treatment with MMF or ABA was more strongly associated with non-response to Sars-CoV-2 vaccination ( Figure 1H).

Antibody Titers
When analyzing antibody titers after vaccination, we found that IRD patients showed overall lower mean titers compared to controls (Figure 2A No differences were identified when patients and controls were stratified by age, group, and sex ( Figures 2B-C). When we looked at the different IRD diseases, CTD, RA, and SpA had lower titer  Figure 2D). Past COVID-19 was associated with higher titers than patients without prior infection ( Figure 2D). CCS users surprisingly showed lower titer compared to patients not under CCS (IQR) 189 (39.7-534.5) vs. 576 (192-1300), p < 0.001; likely due to small sample size (n = 5), and no differences were observed when studying the titer of patients who received RTX in last 12 months ( Figure 2E). Nearly all treatments appeared to influence antibody titers, except for Sulfasalazine, Hydroxychloroquine (HCQ), Colchicine (COL), IL1 inhibitors (IL1i) and Ustekinumab (UST). Treatment with Azathioprine (AZA), MTX, and MMF led to lower titer compared either to controls and other patients not taking any csDMARDs ( Figure 2F

Linear Regression Model
After performing a backward stepwise linear regression model using anti-Sars-Cov-2 titer as the independent variable and including all treatment data, Rituximab, patient age, and sex, we identified that MMF, ABA, CCS, MTX, JAKi, TNFi, age and male sex was associated with antibody titer ( Figure 2H) Figure 2H).

DISCUSSION
Randomized clinical trials and real-life data demonstrated the safety and efficacy of the currently available mRNA vaccine leading to approval nearly worldwide (14,15). The immunological response is commonly altered in IRD patients and it is the target of most of the current treatments (16).
Independent studies following the approval of mRNA vaccines demonstrated their efficacy and safety also in IRD and immunocompromised patients (6). However, a higher percentage of non-responders are observed in immunocompromised patients. A sizable proportion of IRD patients fails to mount a serological response against Sars-Cov-2: in our cohort globally 11.4% had no detectable levels of IgG anti-Sars-Cov-2 with a peak of 56.2% in MMF treated patients, which is in line with other studies (11). Therefore, there is an emerging and urgent need to identify vulnerable, unprotected IRD patients and elaborate strategies aiming to maximize the protection against severe COVID-19 in these patients. This work aimed to identify the cause of the lack of antibody-mediated immune response against SARS-CoV-2 in a cohort of Italian adult IRD patients. To date, this study is among the largest controlled cohorts evaluating the BNT162b2 vaccine in a rheumatologic setting, thus adding a significant contribution to previous similar studies.
As evidenced in our cohort, although the BNT162b2 vaccine showed a good immunogenicity profile according to seroconversion rate, a considerable portion of patients affected by IRD failed to mount an appropriate humoral response. This data seems consistent with evidence shown in other cohorts, with comparatively similar seroconversion rates (11,17,18).
Treatment with CCS, MMF, and ABA were consistently associated with failed seroconversion and lower anti-spike antibody titers, thus suggesting that these medications alone might be responsible for ineffective vaccination.
In the light of the common use of MMF in rheumatology to treat severe systemic conditions, including inflammatory lung disease and glomerulonephritis that are per se associated with a higher risk of death due to COVID-19, this observation appears of high clinical relevance (4,19).
Such associations find confirmation in other studies on anti-SARS-CoV-2 vaccines (11,17,20,21) and mirror what we know from other vaccination data (22)(23)(24)(25)(26).. At the same time, MMF negatively impacted immunogenicity after the HPV vaccine in patients with Systemic Lupus Erythematosus (27) and response to influenza vaccine in kidney transplant recipients (28). In this regard, the evidence that MMF impairs COVID-19 vaccine immune response is widely recognized; and conversely, data on ABA were still considered by some authors and by the EULAR as inconclusive for the risk of bias and the small sample size of the published study (9). However, more data are accumulating (11,13) and the results of this controlled study gains added value as it contributes to the identification of ABA as a negative predictor of response.
Corticosteroids appeared to have either a relatively neutral (29) or a moderately negative influence on seroconversion rates after vaccination against influenza viruses (30,31). Using regression analysis, we tried to pull apart the contribution of CCS, demonstrating a negative effect on both seroconversion rate and antibody titer. One of the limitations of our study is the stratification of patients in CCS users versus non-CCS users, not taking into consideration the prednisone equivalent dose in the analysis. This choice was driven by the need for an adequate sample size in each group and allowed CCS identification as a factor associated with lower antibody production. The effect size will likely be even more prominent in patients taking high doses of corticosteroids.
An unusual finding was the relatively modest effect of Rituximab on seroconversion rates in our cohort. As evidenced in other studies on COVID-19 vaccines and data from studies on influenza and pneumococcal vaccines, Rituximab exerts a potent action on antibody production and is known to compromise immunogenicity of COVID-19 vaccine, as well as other vaccines (6,(32)(33)(34). Furthermore, B cell depleting therapy seems to influence also the T cells function by compromising the circulating T follicular helper cell response and augmenting the CD8 T cell induction (35).
However, the patients treated with RTX in the last 12 months were only five in our cohort.
The negative effect exerted by MMF, ABA, CCS, MTX, JAKi, RTX, and TNFi is partially in line with previously reported information (2,7,21,(36)(37)(38) and it is of relevance independent from the definition of seroconversion. Although a clear cut-off of antibody titer associated with protection is lacking, the data available demonstrates that the protection drops during subsequent months (39) and this supports the policy of administering a third and fourth booster dose (40)(41)(42) and maybe more in the future. Immunosuppressive therapy may lead to a faster loss of protection in IRD patients. In this context, the timing of testing chosen in this study may be more informative. One limitation of the study is the lack of longitudinal data; however, testing two months (8 weeks) after the second dose may allow catching a steady state.
One criticism against serological studies is the lack of information about cell-mediated immunity, particularly T-cell response and to the neutralization of the emerging variants, including the Omicron. Since T-cell immunity may confer protection also in the absence of antibodies (43)(44)(45), antibody titer may be only the tip of the iceberg (46)(47)(48)(49). Most commonly, drugs such as ABA and MMF impair both T cells and B cells' (50) mediated response, possibly exposing the patients to a higher susceptibility to severe COVID-19. A low serological response may well indicate a poor response and protection in these cases.
At the moment, the accumulating evidence on the impact of these treatments on vaccine immunogenicity should prompt a reflection about whether it would be advisable for patients to temporarily stop therapy with MMF and ABA and at least reduce daily corticosteroid dose before vaccination. Indeed, data from a small cohort suggests that adopting a temporary hold policy for MMF was associated with higher anti-SARS-CoV-2 antibody titers compared to patients who did not withhold the medication (92% vs 65% seroconversion rate (51). However, this strategy is still not backed up by solid evidence and further studies are required to establish an exact timing for the temporary hold of each treatment to achieve an adequate balance between vaccine efficacy and disease control.
In this context, clinical guidance and recommendations on the COVID-19 vaccination for patients with rheumatic and musculoskeletal disease has been released and recently updated by the American College of Rheumatology (ACR) and the European Alliance of Associations for Rheumatology (EULAR) (9,52).
The decision to whether temporarily withhold a treatment for IRD is particularly delicate and requires reliable scientific data to perform a risk-benefit evaluation. The temporary interruption of the treatment may expose the patients to a higher risk of flare and needs to be balanced with the risk of severe COVID-19.
Finally, a universally accepted cut-off for the serological antibody titer offering protection to severe COVID-19 has not been identified, however it appears relevant that a proportion of patients did not have any serological response, and these are likely to be at higher risk. Seroconversion rates among patients with autoimmune and inflammatory rheumatic disease are lower overall compared to controls, and this effect is imputable to concomitant therapy.
Treatment with corticosteroids, MMF, ABA, and older age are associated with the absence of serological response to the BNT162b2 mRNA anti-SARS-CoV-2 vaccine. Treatment with MMF, ABA, CCS, MTX, JAKi, TNFi, and RTX reduces the antibody titer and may accelerate the drop in immunity against SARS-CoV-2.
Shared evidence-based strategies (51) should be developed to identify patients at high risk of being unprotected and improve their response to SARS-CoV-2 vaccines.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by Ethical Committee of the University of Campania L. Vanvitelli. The patients/participants provided their written informed consent to participate in this study.

AUTHOR CONTRIBUTIONS
DM, AC, CDV, and FC developed the study protocol. DM, AC, CDV, GR, MDP, CC, FR, SS, MP, and LB contributed to enrollment, sample collection, data collection and data quality control. MV and LP performed serological analysis. DM performed data analysis and interpretation and prepared the figures. DM and AC drafted the manuscript. All the authors contributed to interpreting the data and revised the manuscript critically for important intellectual content.